Interpreting In/out referred noise of chopper amplifier

[cl254]

Hi Everyone,

After simulating a chop amp, it seems like looking at the input referred noise, the 1/f noise is moved to the modulation freq (clk freq), but when I look at the output referred noise, I still see 1/f noise at DC level. I thought the 1/f noise should have been modulated up by the demodulator (the second modulator) when I look at output referred noise.

The input referred noise looks right because the noise at output is divided by the gain, so the shifted 1/f appears right.

Have I interpreted "output referred noise" wrong?

Thanks.

 

[FvM]

Apparently, some noise sources are located outside the modulated/demodulated signal path.
You should be able to identify them in the simulation.

 

[cl254]

Hi FvM,

Thanks for the reminder, actually I left the output of demod with no connection and still see 1/f noise at DC!

Now I suspect that Cadence uses another set of output noise other than the one it uses to calculate input referred noise, this is the most reasonable answer, but still does not make sense.

 

[cl254]

I think now I can close my own loop.

My output is differential, there is a R and a C in parallel between the outputs. I measured between the outputs for noise simulation.

With the RC, the output referred noise has 1/f^2 component, and the drop is 20dB/dec instead of 10dB/dec.

Without the RC, the output referred noise is correct.

So basically the computation of output referred noise is related to the parallel RC and somehow brought in 20dB/dec of noise drop at 1kHz.

If anyone knows how this happens, please let me know.